New Data Structure for Computational Molecular Design with Atomic or Fragment Resolution

Abstract

A new molecular data structure and molecular structure operation algorithms are proposed for general purpose molecular design. The data structure allows for a variety of molecular operations for creating new molecules. Two types of molecular operations were developed, unimolecular and bimolecular operations. In unimolecular operations, a child molecule can be created from a parent via addition of a functional group, deletion of a fragment, mutation of an atom, etc. In bimolecular operations, children molecules are generated from two parent molecules through combination or crossover (hybridization). These molecular operations are essential for the creation and modification of molecules for the purpose of molecular design. The data structure is capable of representing linear, branched, multifunctional, and multivalent compounds. Algorithms are developed for deriving the molecular data structure of a molecule from its atomic coordinates and vice versa. We show that this new molecular data structure and the developed algorithms, referred to as Molecular Assembling and Representation Suite, allow one to generate a comprehensive library of new molecules via performing every possible molecular structure modification.

MARS Plus: An Improved Molecular Design Tool for Complex Compounds Involving Ionic, Stereo, and Cis–Trans Isomeric Structures

“…43 In our previous work, we have proposed a graph-like 2D molecular data structure (MDS) with atom-based genetic algorithms (i.e., crossover, mutation, and combination) included for the design of specialty chemicals, referred to as the Molecular Assembling and Representation Suite or MARS. 1 The MDS consists of integer arrays, and new molecular structures can be generated through the application of genetic operations on the integer arrays, which can then be converted to 1D SMILES or 3D coordinate representations. Despite its success, the MDS of MARS was limited to simple neutral molecules and cannot be applied to design charged species or isomers with cis−trans isomerism and enantioisomerism.…”